def getMP(): MP = [] from htmldecode import htmlentitydecode, set_pretty_formatting, set_quote_and_code_in_color #html = open( os.path.join( CACHEDIR, "pm.html" ), "r" ).read() html = get_page( "http://passion-xbmc.org/pm/", filename='pm.html' ) windowbg = re.compile( '<td colspan="2" class="windowbg(.*?)<td valign="bottom" class="smalltext" width="85%">', re.DOTALL ).findall( html ) for mp in windowbg: auteur = re.findall( '<a href="http://passion-xbmc.org/profile.*?" title="Voir le profil.*?" style="color\:#(.*?)\;">(.*?)</a>', mp ) auteur_avatar = re.findall( '<img src="(.*?)" alt="" class="avatar reflect" border="0" />', mp ) if auteur_avatar: auteur_avatar = get_MP_avatar( auteur_avatar[ 0 ] ) else: auteur_avatar = "" title, date = re.findall( '<td align="left" valign="middle">.*?<b>(.*?)</b>.*?<b> le:</b>(.*?) »</div>', mp, re.DOTALL )[ 0 ] isread = re.findall( '<div class="smalltext">« (.*?) »</div>', mp, re.DOTALL ) message = re.findall( '<div class="personalmessage">(.*?)</div>\n', mp, re.DOTALL ) auteur = "[COLOR=FF%s]%s[/COLOR]" % ( auteur[ 0 ][ 0 ], auteur[ 0 ][ 1 ], ) title = set_pretty_formatting( htmlentitydecode( unicode( title, 'utf-8' ) ) ) date = set_pretty_formatting( htmlentitydecode( unicode( date, 'utf-8' ) ) ) if len( isread ) >= 2: isread = htmlentitydecode( unicode( isread[ 1 ], 'utf-8' ) ) else: isread = "" message = htmlentitydecode( decode_smileys( set_pretty_formatting( set_quote_and_code_in_color( unicode( message[ 0 ], 'utf-8' ) ) ) ) )#+"</div>" #le reste du nettoyage ce passe dans GUI/ForumDirectInfos.py #message = re.sub( "(?s)<[^>]*>", "", message ) MP.append( ( auteur_avatar, auteur, title, date, isread, message ) ) return MP
def getMP(): MP = [] from htmldecode import htmlentitydecode, set_pretty_formatting, set_quote_and_code_in_color #html = open( os.path.join( CACHEDIR, "pm.html" ), "r" ).read() html = get_page("http://passion-xbmc.org/pm/", filename='pm.html') windowbg = re.compile( '<td colspan="2" class="windowbg(.*?)<td valign="bottom" class="smalltext" width="85%">', re.DOTALL).findall(html) for mp in windowbg: auteur = re.findall( '<a href="http://passion-xbmc.org/profile.*?" title="Voir le profil.*?" style="color\:#(.*?)\;">(.*?)</a>', mp) auteur_avatar = re.findall( '<img src="(.*?)" alt="" class="avatar reflect" border="0" />', mp) if auteur_avatar: auteur_avatar = get_MP_avatar(auteur_avatar[0]) else: auteur_avatar = "" title, date = re.findall( '<td align="left" valign="middle">.*?<b>(.*?)</b>.*?<b> le:</b>(.*?) »</div>', mp, re.DOTALL)[0] isread = re.findall('<div class="smalltext">« (.*?) »</div>', mp, re.DOTALL) message = re.findall('<div class="personalmessage">(.*?)</div>\n', mp, re.DOTALL) auteur = "[COLOR=FF%s]%s[/COLOR]" % ( auteur[0][0], auteur[0][1], ) title = set_pretty_formatting(htmlentitydecode(unicode(title, 'utf-8'))) date = set_pretty_formatting(htmlentitydecode(unicode(date, 'utf-8'))) if len(isread) >= 2: isread = htmlentitydecode(unicode(isread[1], 'utf-8')) else: isread = "" message = htmlentitydecode( decode_smileys( set_pretty_formatting( set_quote_and_code_in_color(unicode( message[0], 'utf-8'))))) #+"</div>" #le reste du nettoyage ce passe dans GUI/ForumDirectInfos.py #message = re.sub( "(?s)<[^>]*>", "", message ) MP.append((auteur_avatar, auteur, title, date, isread, message)) return MP
def _decode( text ): try: return htmlentitydecode( text ) except: print_exc() return text
def _decode(text): try: return htmlentitydecode(text) except: print_exc() return text